Now, I want to talk about scatter plots and faceting. So in a scatter plot, if we want to visualize an additional attribute, one channel that we can use is color. So what you see here, is another scatter plot similar to what we have seen before, but with the addition that now there is an additional attribute, a third attribute, that encodes information about a number of categories, each category, one color. The interesting thing of scatter plots is that you can add even more attributes if you want to. So in this case, we have exactly the same scatter plot, which is encoding one attribute on the X-axis, one on the Y-axis, now another one with color, and even another one with the size of the dots or bubbles. A scatter plot can encode up to four attributes. In general, you could add even more channels and attributes, but the more you add, the harder it's going to be to actually decode the information that is contained in our scatter plot. Now, I want to show you what you can do with exactly the same scatter plot. Look at this. Starting from these specific scatter plot, what we can do is to create a replica of the same scatter plot, where every single one contains only the data of one of the categories. Here we have one, two, three, four, five scatter plots. They all come from the original scatter plot, and each one shows only the data that belongs to one of the categories that you see in the region on the right. This is a very general purpose technique, that is called faceting, and visualization method that you see here is also sometimes called small multiples. Why small multiples? Well, because you get the same graph and you create multiple replications of the same graph, where each one contains a subset of the data, that's why it's called small multiples. The interesting thing about this technique, is that it can be applied to any graph. In this case, I've shown you an example of faceting with scatter plots but, as I'm going to show you in a moment, you can apply exactly the same method to virtually any other graph. How does faceting work? You select one categorical or ordinal attribute, then you create as many sets as the number of values that you have in this attribute, and then you create one plot for each of these values. Here is how it looks like for the scatter plot that we have just seen. You start from the original scatter plot, you take every single category that comes from one selected attribute, and you multiply this information, showing the information of only one category at the time, in each scatter plot. The same thing, as I said, can be done with virtually any other graph. Here is a different example. Faceting with maps. I have information, I have data about the city of New York, and I split this information into different categories, and each map represents the data coming only from one of the categories. Exactly the same thing can be done, for instance, with line charts. Here is another example, I have one, two, three, four,five different categories of things that are changing over time, and rather than putting them all together in one graph overlapping, I create one graph for each of them. As I said, this is a very general purpose technique, and it's incredibly useful in many situations, especially when you have too many categories to show at once, and you don't have an easy way to put all these categories together in one single plot. Here, you can just split them into a number of plots, and create what is called a small multiples presentation.